|Publication number||US7028022 B1|
|Application number||US 09/627,662|
|Publication date||Apr 11, 2006|
|Filing date||Jul 28, 2000|
|Priority date||Jul 29, 1999|
|Also published as||CA2279119A1, CA2279119C|
|Publication number||09627662, 627662, US 7028022 B1, US 7028022B1, US-B1-7028022, US7028022 B1, US7028022B1|
|Inventors||Sam S. Lightstone, Catherine S. McArthur, William T. O'Connell, Miroslaw A. Flasza|
|Original Assignee||International Business Machines Corporation|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (24), Non-Patent Citations (9), Referenced by (27), Classifications (13), Legal Events (5)|
|External Links: USPTO, USPTO Assignment, Espacenet|
This patent application claims priority from the commonly assigned Canadian Patent Application entitled “Heuristic-Based Conditional Data Indexing”, having Canadian Patent Application Serial No. 2,279,119, filed on Jul. 29, 1999 by Sam S. Lightstone, Miroslaw A. Flasza; Catherine S. McArthur and William T. O'Connell, which application is incorporated herein by reference in its entirety.
1. Field of the Invention
The present invention is directed to an improvement in computing systems and in particular to computer systems which include the indexing of data.
2. Background of the Art
Computer systems for the storage and retrieval of data typically include indexes of the data which are created to permit operations on the data (data retrieval or updating) to be performed in an efficient manner. Updating such indexes is typically costly in system resources and time. For example, in a relational database environment, appending a new record to a table in the database will require a new entry in any indexes which have been created for that table. The time and computer resources required to add a set of data to the table may be significantly less than the operation of updating the associated indexes.
In database management system (DBMS) environments, it is typically the case that the efficiency or inefficiency of the index updating operations significantly impacts on the efficiency of the database operations as a whole.
It is therefore desirable to have a DBMS computer system which includes features to make more efficient the updating of indexes associated with data stored by the DBMS.
According to one aspect of the preferred embodiments, there is provided an improved system for indexing data.
According to another aspect of the preferred embodiments, there is provided an improved method, system, and program for updating an index on a database table when data is added to the table. When data records are received to load into the table, a selection is made of a first operation or second operation. The first operation incrementally updates the index on the table as each record in the received data is added to the table and the second operation rebuilds the index from the table after all the received data records have been added to the table. The selected first or second operation is used to update the index with the received data.
In still further embodiments, a determination is made as to which of the first operation or second operation is more efficient. The first or second operation determined to be more efficient is the selected operation used for updating the index with the received data.
According to another aspect of the preferred embodiments, there is provided a computer system for database management comprising, means for storing and updating a first set of data, indexing means for storing and updating a selected index of key values related to the first set of data, the indexing means comprising means for selectively updating the index by incrementally updating the index by incrementally adding key values to the index or by fully rebuilding the index, heuristic determination means for selecting the incremental update of the index, or the full rebuild of the index, for a given second set of data to be added to the first set of data.
According to another aspect of the preferred embodiments, there is provided the above computer system in which the heuristic determination means comprises a function which takes as input index meta-data, comprising characteristics of the first set of data, the index, and the second set of data.
According to another aspect of the preferred embodiments, there is provided the above computer system in which the database management system is a relational database management system in which the index is stored as a binary tree and in which the index meta-data comprises estimates of the table size of the first set of data, the table size of the second set of data, and the height of the index, whereby the function is empirically defined to select the incremental update of the index or the full rebuild of the index based on the predicted relative efficiencies of the incremental update of the index and the full rebuild of the index.
According to another aspect of the preferred embodiments, there is provided the above computer system in which the function determines, for each potential binary tree height of the index, a threshold percentage of the table size of the second set of data to the table size of the first set of data, whereby the threshold percentage value for each binary tree height determines the selection of the incremental index update or the full rebuild index update.
According to another aspect of the preferred embodiments, there is provided the above computer system in which the heuristic determination means comprises means for the user to specify the selection of the incremental update of the index or the full rebuild of the index.
According to another aspect of the preferred embodiments, there is provided the above computer system further comprising means for generating values for the function of the heuristic determination means, for a specified test range of values for each of the first set of data, the index, and the second set of data.
According to another aspect of the preferred embodiments, there is provided the above computer system, the computer system having one or more CPUs, one or more disks, a sort heap and a database bufferpool, and in which the index meta-data is defined to reflect a subset of the following characteristics: the percentage of free space in the index, the estimated size of the index after both the incremental and the rebuild updates of the index, the width of the average key value in the index, the size of the sort heap and the database bufferpool in the computer system, the number and speed of the CPUs in the computer system, and the number and speed of the disks in the system.
According to another aspect of the preferred embodiments, there is provided a computer system for relational database management comprising, means for storing and updating a first set of data, indexing means for storing as a binary tree, and updating, a selected index of key values related to the first set of data the indexing means comprising means for selectively updating the index by incrementally updating the index by incrementally adding key values to the index or by fully rebuilding the index, heuristic determination means for selecting the incremental update of the index, or the full rebuild of the index, for a given second set of data to be added to the first set of data, the heuristic determination means comprising a function which takes as input index meta-data, the index meta-data being stored separately from the first set of data and comprising estimates of the table size of the first set of data, the table size of the second set of data, and the height of the index, whereby the function is empirically defined to select the incremental update of the index or the full rebuild of the index based on the predicted relative efficiencies of the incremental update of the index and the full rebuild of the index, the heuristic determination means further comprising means for the user to specify the selection of the incremental update of the index or the full rebuild of the index.
According to another aspect of the preferred embodiments, there is provided a computer program product for use with a computer comprising a central processing unit and random access memory, said computer program product comprising a computer usable medium having computer readable code means embodied in said medium for managing a database, said computer program product comprising computer readable program code means for causing a computer to implement the system functionality described above.
According to another aspect of the preferred embodiments, there is provided a method for updating a selected index in a computer system for database management, the computer system comprising means for storing and updating a first set of data related to the index, the method comprising the steps of:
According to another aspect of the preferred embodiments, there is provided the above method in which the step of heuristically selecting the update of the index comprises the evaluation of a function which takes as input index meta-data, comprising characteristics of the first set of data, the index, and the second set of data.
According to another aspect of the preferred embodiments, there is provided a method for updating a selected b-tree index in a computer system for relational database management, the computer system comprising means for storing and updating a first set of data related to the index, the method comprising the steps of:
According to another aspect of the present invention, there is provided the above method in which the function comprises the step of determining, for each potential binary tree height of the index, a threshold percentage of the table size of the second set of data to the table size of the first set of data, whereby the threshold percentage value for each binary tree height determines the selection of the incremental index update or the full rebuild index update.
According to another aspect of the present invention, there is provided the above method further comprising the step of a user specifying the selection of the incremental update of the index, the full rebuild of the index, or the selection based on the heuristic determination function.
According to another aspect of the present invention, there is provided a computer program product tangibly embodying a program of instructions executable by a computer to perform the above method steps.
Advantages of the present invention include improvements in the efficiency in updating a database when loading data or when rolling forward data after a database restore.
The preferred embodiment of the invention is shown in the drawings, wherein:
The preferred embodiment is described with reference to the DB2 database management system (DBMS).** Those skilled in the art will appreciate that the preferred embodiment may be implemented in other DBMS environments other than DB2. The preferred embodiment of the invention relies on the fact that for some database operations a series of transactions will result in data being updated without there being a need to access indexes associated with the data.
**DB2 is a registered trademark of International Business Machines Corp.
For example, a DBMS may permit a user to load data into a table in the database from another source (for example where data is copied from a first database into a second database). The DB2 LOAD command is an example of this sort of operation. The LOAD command will result in data being inserted into a table. It is possible for large data collections to be added to tables using the LOAD command. In one type of prior art system, the DBMS will incrementally update indexes associated with the table into which data is placed using the LOAD command. In such a system, the index on the table will be modified on a record by record basis to reflect the new records added to the table by the LOAD command. However, the index will not be used, in the sense of being accessed to locate data in the database, until the LOAD command has been completed and all the data has been added to the target table in the database. In other systems, the index will not be updated in an incremental fashion but will be fully rebuilt following the addition of new records to the data in the database.
Another example of where a DBMS adds to or modifies data without the use of an index until the changes to the data are complete is in the ROLLFORWARD command in DB2. When data in a relational database is recovered by using the RESTORE and ROLLFORWARD commands, the DBMS facilitates the rebuilding of a database from a transaction log where the database has been destroyed or become corrupted in some manner (a database “roll forward”). During the roll forward process, the data in the database is modified to reflect the transactions stored in the transaction log. In the prior art, it is common to update associated indexes on a record by record basis which reflects the transactions stored in the transaction log. For DBMS environments such as the DB2 system, the transaction log stores a portion of the transactional history which permits the updating of the data in the database after a database restore, without requiring reference to the indexes associated with the data being updated or rolled forward. The indexes associated with the data must be brought up to date at the end of the roll forward operation, but need not be synchronized with the data during the operation.
As indicated above, for certain DBMS environments, the DBMS will update indexes on a transaction by transaction or record by record basis. For certain DBMS environments, the content of the indexes during the data updating process (during the LOAD or the roll forward command execution) is not accessed by the DBMS. The DBMS requires a correct set of indexes at the completion of the operations but is not concerned with the intermediate values of the indexes.
As the above indicates, in the prior art, one approach in loading data or performing a roll forward on a database, is to update the database on a record by record basis. In such a system any indexes which are associated with the data being updated will also be modified on a record by record basis. Alternatively, certain DBMS designs will permit indexes to be rebuilt at the end of a roll forward operation. In certain DBMS designs, it is also possible for the system itself to select an index rebuild in certain circumstances where otherwise a record by record or incremental update of the index would occur. In the DB2 DBMS, for example, the default approach during the ROLLFORWARD command execution is the incremental updating of the index. The index rebuilding operation is only carried out where one of the transactions being replayed in a roll forward is an index create function. Otherwise, the entire set of transactions is replayed on the index.
Where data is loaded or a database is subject to a roll forward operation, it is possible that a significant proportion of the processing time for the operation will be taken up by the addition to or modification of the indexes associated with the data. In fact, for certain databases it may be more efficient to ignore the incremental updating of the associated indexes and to fully rebuild the indexes at the completion of the loading of data or the roll forward operations.
In the preferred embodiment, a backup of the database is made at periodic or user-defined intervals. In
According to the preferred embodiment, the DBMS includes functionality to recover data using commands such as RESTORE and ROLLFOWARD, which does not necessarily require the record by record updating of index 14. In the preferred embodiment, backup database 18 may be used to replace the table 12 in database 10. However, when the transactions in log 24 are replayed on database 10 as in a roll forward, the system of the preferred embodiment will selectively permit table 12 to be updated, without updating index 14 (as is set out above, log 24 contains data for each of the stored transactions to permit updating of table 12 without reference to index 14). As the time for incrementally updating index 14 is potentially long, the savings in updating table 12, only are potentially significant. Once table 12 has been updated by the transactions in log 24, index 14 will be recreated by being rebuilt on the newly updated table 12.
Such a method of updating by ignoring index 14 until the completion of updating table 12 (the full index rebuild) is available in the preferred embodiment by user selection. Alternatively, the full index rebuild roll forward is available by the system of the preferred embodiment applying a heuristic determination function to meta-data 16.
Index meta-data 16 is represented as Si. A heuristic determination function F(Si) is defined in the system of the preferred embodiment to accept the set of heuristic information and to determine whether indexing in a roll forward should be maintained transactionally (incremental updating), or deferred for a single recreation pass (full index rebuild).
The determination function F(Si) may accept as parameters such database meta-data as the estimated number of key-parts assumed to comprise an index at the time roll forward processing starts, and an estimate of the number of keypart updates that will be applied during roll forward processing. The specific decision making logic of the heuristic determination function F(Si) may be dependent on implementation details, for specific DBMS systems. The F(Si) definition may have considerable empirical dependence, which may vary between products, and between product releases.
A simple implementation of function F(Si) has parameters reflecting characteristics of an index 22, and compares these to a set of index characteristics at roll forward completion time which have been maintained in meta-data 16. In the specific implementation for the DB2 Universal Database, the preferred embodiment stores the number of index key operations as meta-data inside the index object meta-index. This information is periodically reflected in the database recovery history file during normal transaction processing of the DBMS. In the preferred embodiment this information is stored outside the database to maintain the information should the database crash. During roll forward recovery processing the desired heuristic information reflecting end-of-logs can be retrieved from the Recovery History File, and provided to the heuristic determination function F(Si).
In the system of the preferred embodiment, the heuristic determination function F(Si) is used also when pre-existing data is loaded into a database.
The implementation of the loading of data in the preferred embodiment system may be described with reference to FIG. 3. The first step in loading data 26 is the addition of the data to table 12. The added data 38 shown in table 12 in
As is apparent, in both the data recovery and roll forward operations, and in the data load operation, the use of the heuristic determination function F(Si) is available in the system of the preferred embodiment. The function will be used (FIG. 4: reference numeral: 38. 40. 42. 44) to select (FIG. 4: reference numeral 40) incremental updating (FIG. 4: reference numeral 42) index 14, or alternatively to select a full rebuilding (FIG. 4: reference numeral 44) of index 14. In the system of the preferred embodiment, the user may either manually select incremental or full rebuilding modes or may system option to have the system itself use the heuristic determination function F(Si) to make the selection.
In both the load and restore/roll forward operations, the heuristic determination function F(Si) is intended to predict which of the two index updating approaches (incremental and full rebuild) will be most efficient for the database updating which is being performed. The case where database 10 is relatively large, and few data values are added to database 10 (and correspondingly to index 14), will be, for many DBMS environments, one where incremental updating of index 14 will be more efficient. Conversely, where database 10 is relatively small and the data to be added by either a large log 24 or a large set of data 26, in many DBMS environments, it will be more efficient to update index 14 by fully rebuilding the index after table 12 has been updated. Typically, it is the database subsystem and environment which will determine which of the two different updating methods is more efficient, for a given data and index updating operation. The heuristic determination function F(Si) is intended to predict which of the two methods is more efficient, given a particular combination of existing database, update data, database subsystem and environment. Once the appropriate values for the function are set for a given database subsystem and environment, it is intended that the same function F(Si) will be appropriate for different table and index combinations in different database definitions.
In the system of the preferred embodiment, index 14 is stored in the form of a binary tree, such as a B-tree, B+-tree, or B*-tree. It is to be understood, however, that other index structures may be used and the benefit of the invention achieved. The preferred embodiment uses two characteristics of a binary tree index to model the efficiency of incremental updating and full rebuilding for an index. The two characteristics are the percentage of new data that is to be added to table 12 and the height of the binary tree structure of index 14. The tree height of index 14 is a measure of the size and complexity of index 14. In the example of the preferred embodiment, there is a single index shown. In the system of the preferred embodiment, the first table non-meta-index table index is taken to be representative of the database indexes which will be affected by the updating procedure. It will be understood by those skilled in the art that there may be other approaches to selecting an index which is representative of the indexes in a database for which data is to be updated. The approach of selecting the first such index and assuming that that index is representative has the advantage of carrying low storage and processing overheads. It may be advantageous in certain circumstances to maintain other data regarding the indexes such as the largest index, the smallest index, or the median or average index. Such approaches may require more system resources to maintain but will have the advantage of potentially providing better information about the structure of the indexes for the determination heuristic F(Si).
In defining F(Si) in terms of tree height of index 14, and the percentage of new data to be added to table 12, F(Si) is designed to reflect the time required to extract index keys from table 12. This operation is necessary in a full rebuild of index 14 and requires a scan of all records in table 12. The cost of this is generally linear with the size of the data on disk. Typically, each page must be scanned from disk, then scanned for records, and each record scanned for index keys.
The function F(Si) is also designed to reflect the time to sort index keys. This varies with the sorting algorithm used, the size of buffering for sort operations, and the cost of spilling to disk (if required). Generally, however, this cost is modelled as an NlogN operation.
The function is also intended to reflect the time to fully rebuild index 14 from a set of sorted keys. This is generally a linear operation, which varies with the number of key values to be included in index 14. Conversely, the function will be used to reflect the time to incrementally extend index 14 with a set of sorted keys. This is generally a non-linear operation, which is a function of the height of the binary index tree.
In the model for F(Si) of the preferred embodiment, a threshold value for percentage of new data is determined for different tree heights. The threshold value will indicate at what point it is estimated that incremental updating of index 14 is to be used (the threshold value and below) and at what point a full rebuild approach to index 14 should be used (above the threshold value).
In the preferred embodiment, the F(Si) is essentially a table:
2 (or less)
In the table T2 . . . T8 represent the percentile thresholds for the tree height 2 to 8 for which incremental indexing proves to yield superior performance for a given DBMS. As the tree height grows, incremental indexing threshold values will vary. The incremental costs of sorting data need to be balanced against the cost of traversing the tree. Since both sorting and scanning the binary tree vary as NlogN operations, to some extent the way these two factors trade off with varying amounts of new and old data depends on the implementation of the key sorting and tree insertion logic. Due to the dependency on implementation, the preferred embodiment models these costs empirically. Tests are performed to examine at what volumes of new index key values, incremental indexing performed better than full index rebuilding.
In the preferred embodiment, the F(Si) for the DB2 product was determined by using as benchmark data a set of data provided by the Transactional Processing Performance Counsel. Based on that benchmark data, the values for the thresholds for tree heights 2 to 8, inclusive, for the DB2 product were calculated as:
Additional values may be calculated for tree heights greater than eight, as required. The above values indicate that where index 14 has a height 6, for example, and the data to be added to table 12 is above 27% of the size of table 12 (the number of records to be added is more than 27% of the number of records in table 12), then a full rebuild of index 12 will be more efficient than adding key values to index 12 incrementally. It is to be understood that the above values may vary for different DBMS environments and data sets.
Although the above table values for the function F(Si) were arrived at by taking test data and manually constructing indexes to determine the thresholds for the different heights of trees, such a process need not be carried out by a user. Once the location of the test data is indicated to the system of the preferred embodiment, it is possible for the system to run the tests to come up with the appropriate threshold values as part of the system set up. It is necessary for the system to clock its own processes and to select the appropriate threshold points based on the data provided.
For indexes such as hash indexes which are not based on a binary tree structure, the heuristic determination function may compare percentages of new data to total data volume to achieve a heuristic to predict when incremental or full rebuild approaches are more efficient.
Other characteristics of database 10 which may be included in determining the function F(Si) include:
Each of the above characteristics may have an impact on the relative efficiency of incremental and full rebuild updating of index 14. For that reason, inclusion of these values in the function F(Si) may provide additional predictive accuracy for the function.
Although a preferred embodiment of the present invention has been described here in detail, it will be appreciated by those skilled in the art, that variations may be made thereto, without departing from the spirit of the invention or the scope of the appended claims.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US4648031||Jun 21, 1982||Mar 3, 1987||International Business Machines Corporation||Method and apparatus for restarting a computing system|
|US5095423 *||Mar 27, 1990||Mar 10, 1992||Sun Microsystems, Inc.||Locking mechanism for the prevention of race conditions|
|US5129082 *||Mar 27, 1990||Jul 7, 1992||Sun Microsystems, Inc.||Method and apparatus for searching database component files to retrieve information from modified files|
|US5230047||Apr 16, 1990||Jul 20, 1993||International Business Machines Corporation||Method for balancing of distributed tree file structures in parallel computing systems to enable recovery after a failure|
|US5455946||May 21, 1993||Oct 3, 1995||International Business Machines Corporation||Method and means for archiving modifiable pages in a log based transaction management system|
|US5561795||May 13, 1994||Oct 1, 1996||Unisys Corporation||Method and apparatus for audit trail logging and data base recovery|
|US5717921||Mar 17, 1995||Feb 10, 1998||Digital Equipment Corporation||Concurrency and recovery for index trees with nodal updates using multiple atomic actions|
|US5742805||Feb 15, 1996||Apr 21, 1998||Fujitsu Ltd.||Method and apparatus for a single history register based branch predictor in a superscalar microprocessor|
|US5758356||Sep 12, 1995||May 26, 1998||Hitachi, Ltd.||High concurrency and recoverable B-tree index management method and system|
|US5764877||Sep 1, 1995||Jun 9, 1998||Digital Equipment Corporation||Media recovery with time-split B-trees|
|US5918225 *||Mar 20, 1997||Jun 29, 1999||Sybase, Inc.||SQL-based database system with improved indexing methodology|
|US5924088 *||Feb 28, 1997||Jul 13, 1999||Oracle Corporation||Index selection for an index access path|
|US6026406 *||Jun 4, 1997||Feb 15, 2000||Oracle Corporation||Batch processing of updates to indexes|
|US6070158 *||Nov 13, 1997||May 30, 2000||Infoseek Corporation||Real-time document collection search engine with phrase indexing|
|US6154746 *||Apr 22, 1998||Nov 28, 2000||At&T Corp.||High-dimensional index structure|
|US6278992 *||Feb 17, 1999||Aug 21, 2001||John Andrew Curtis||Search engine using indexing method for storing and retrieving data|
|US6334124 *||Oct 1, 1998||Dec 25, 2001||Ventro Corporation||Techniques for improving index searches in a client-server environment|
|US6360228 *||Jun 2, 1999||Mar 19, 2002||Oracle Corporation||Transactional framework for executing statements involving non-native code|
|US6457017 *||May 16, 1997||Sep 24, 2002||Softscape, Inc.||Computing system for information management|
|US6553369 *||Mar 11, 1999||Apr 22, 2003||Oracle Corporation||Approach for performing administrative functions in information systems|
|US6591269 *||May 18, 2000||Jul 8, 2003||Sybase, Inc.||Database system with methodology for online index rebuild|
|US6622142 *||Apr 13, 2000||Sep 16, 2003||Staktek Group, L.P.||Database utilities|
|JPH0566984A||Title not available|
|JPH1011345A||Title not available|
|1||*||"Dynamic Maintenance of Wed Indexes Using Landmarks"-Lipyeow Lim, Min Wang, Sriram Padmanabhan, Jeffrey Scott Vitte and Ramesh Agarwal-ACM-May 20-24, 2003 (pps: 102-111).|
|2||"Dynamic Ordering of Joined Rows Using Fields from Multiple Tables", IBM Technical Disclosure Bulletin, vol. 36, No. 11, Nov. 1993, pp. 363-366.|
|3||*||"Incrememtal Updates of Inverted Lists for text Document Retrieval"-Anthony Tomasic, Hector Garcia-Molina and Kurt Shoen-ACM SIGMOD May 1994, (pps: 289-300).|
|4||*||"Synthetic Workload performance Analysis of Incremental Updates"-Kurt Shoens, Anthony Tomasic and Hector Garcia-Molina ACM SIGIR Aug. 1994, (pps: 329-338).|
|5||C. Mohan, "Disk Read-Write Optimizations and Data Integrity in Transaction Systems Using Write-Ahead Logging", IEEE, 1995, pp. 324-331.|
|6||D.J. Haderle, et al., "Index Utilization Enhancement by Deferring Updates and Retrievals", Research Disclosure, May 1990, p. 14-15.|
|7||J. Andersen, "The New Database Wars", InformationWEEK, No. 611, Dec. 1996, pp. 45-47, 49, 51, 53 (Abstract) p. 7.|
|8||K.S. Pang, et al., "An Efficient Recovery Scheme for Large Data in Multimedia DBMS", Journal of the Korea Information Society, vol. 22, No. 2, Feb. 1995, pp. 206-217 (Abstract) p. 6.|
|9||S.S. Lee, et al., "An Index Recovery Method For Real-Time DBMS in Client-Server Architecture", IEEE, Mar. 1997, pp. 110-117.|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US7188124 *||Nov 19, 2003||Mar 6, 2007||International Business Machines Corporation||Method, system, computer product for recovering dropped database table specifying one or more table spaces, recovering the table space being restored using the dropped table history data structure|
|US7290214 *||Apr 29, 2003||Oct 30, 2007||International Business Machines Corporation||Independent deferred incremental refresh of materialized views|
|US7734618||Jun 30, 2006||Jun 8, 2010||Microsoft Corporation||Creating adaptive, deferred, incremental indexes|
|US7779045||Sep 27, 2007||Aug 17, 2010||Microsoft Corporation||Lazy updates to indexes in a database|
|US7818534 *||May 9, 2007||Oct 19, 2010||Yahoo! Inc.||Determination of sampling characteristics based on available memory|
|US7917499||Jun 30, 2006||Mar 29, 2011||Microsoft Corporation||Updating adaptive, deferred, incremental indexes|
|US8086582 *||Dec 18, 2007||Dec 27, 2011||Mcafee, Inc.||System, method and computer program product for scanning and indexing data for different purposes|
|US8103697 *||Jun 18, 2010||Jan 24, 2012||S. Grants Co., Ltd.||Database index key update method and program|
|US8108355 *||Oct 27, 2006||Jan 31, 2012||Hewlett-Packard Development Company, L.P.||Providing a partially sorted index|
|US8239389 *||Sep 29, 2008||Aug 7, 2012||International Business Machines Corporation||Persisting external index data in a database|
|US8381302 *||Sep 14, 2009||Feb 19, 2013||Symantec Corporation||Systems and methods for translating non-comparable values into comparable values for use in heuristics|
|US8527471||Dec 27, 2010||Sep 3, 2013||Sap Ag||Shadow system mirroring of an original system during uptime of an upgrade process|
|US8671087||Dec 5, 2011||Mar 11, 2014||Mcafee, Inc.||System, method and computer program product for scanning and indexing data for different purposes|
|US8924350||Jun 27, 2013||Dec 30, 2014||Sap Se||Shadow system mirroring of an original system during uptime of an upgrade process|
|US8954407 *||Apr 12, 2011||Feb 10, 2015||International Business Machines Corporation||System and method for partially deferred index maintenance|
|US8984514||Nov 30, 2010||Mar 17, 2015||Sap Se||Modifying scheduled execution of object modification methods associated with database objects|
|US9092474 *||Oct 12, 2010||Jul 28, 2015||Sap Se||Incremental conversion of database objects during upgrade of an original system|
|US20040103109 *||Nov 19, 2003||May 27, 2004||Huras Matthew Albert||Dropped database table recovery|
|US20040122828 *||Apr 29, 2003||Jun 24, 2004||Sidle Richard S.||Independent deferred incremental refresh of materialized views|
|US20090193406 *||Jul 30, 2009||James Charles Williams||Bulk Search Index Updates|
|US20120016851 *||Jan 19, 2012||International Business Machines Corporation||System and Method for Partially Deferred Index Maintenance|
|US20120089625 *||Oct 12, 2010||Apr 12, 2012||Sap Ag||Incremental conversion of database objects during upgrade of an original system|
|US20120166404 *||Dec 28, 2010||Jun 28, 2012||Oracle International Corporation||Real-time text indexing|
|US20120173488 *||Dec 29, 2010||Jul 5, 2012||Lars Spielberg||Tenant-separated data storage for lifecycle management in a multi-tenancy environment|
|US20120259824 *||Jun 20, 2012||Oct 11, 2012||International Business Machines Corporation||Maintaining index data in a database|
|CN102542057A *||Dec 29, 2011||Jul 4, 2012||北京大学||High dimension data index structure design method based on solid state hard disk|
|CN102542057B||Dec 29, 2011||Oct 16, 2013||北京大学||High dimension data index structure design method based on solid state hard disk|
|U.S. Classification||1/1, 707/E17.005, 707/999.002, 707/999.003, 707/999.102, 707/999.104|
|Cooperative Classification||Y10S707/99933, Y10S707/99932, Y10S707/99945, Y10S707/99943, G06F17/30336|
|Nov 17, 2000||AS||Assignment|
Owner name: INTERNATIONAL BUSINESS MACHINES CORPORATION, NEW Y
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:LIGHTSTONE, SAM S.;MCARTHUR, CATHERINE S.;O CONNELL, WILLIAM T.;AND OTHERS;REEL/FRAME:011309/0227
Effective date: 20001031
|Jul 17, 2009||FPAY||Fee payment|
Year of fee payment: 4
|May 21, 2010||AS||Assignment|
Owner name: TREND MICRO INCORPORATED,JAPAN
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:INTERNATIONAL BUSINESS MACHINES CORPORATION;REEL/FRAME:024424/0157
Effective date: 20100331
|Oct 16, 2013||FPAY||Fee payment|
Year of fee payment: 8
|Oct 16, 2013||SULP||Surcharge for late payment|
Year of fee payment: 7